Introducing AI assistants with memory

The simple act of remembering is the crux of how we navigate the world: it shapes our experiences, informs our decisions, and helps us anticipate what comes next. For AI agents like Comet Assistant, that continuity leads to a more powerful, personalized experience.

Today we are announcing new personalization features to remember your preferences, interests, and conversations. Perplexity now synthesizes them automatically like memory, for valuable context on relevant tasks. Answers are smarter, faster, and more personalized, no matter how you work. 

How Perplexity remembers

The best AI assistants function as a seamless second brain and operator. Yet, traditional LLMs often break your flow state, hitting context limits that force a restart exactly when you’re most productive. While expanding context windows have helped, they still require “context engineering,” manually summarizing details just to keep the agent on track. Memory solves this by automatically pre-loading your most critical context, ensuring continuity without the manual effort.

With our upgraded personalization, Perplexity remembers key details across conversations so future answers are more personalized and efficient. Structured preferences, like favorite brands, dietary needs, or keywords you often ask about, are securely stored for valuable context. As you ask more of Perplexity, the system builds a dynamic understanding.

A more precise agent 

Most AI chatbots can assemble a user’s history and preferences, but they use that information like they do any other training data: choosing the most probable answer based on language patterns. The history is yours, but the output is an educated guess.

On Perplexity, history is more than a dataset. Instead of treating your history as training data, Perplexity retrieves relevant context from your memory store and uses it directly in crafting better, more accurate answers.

This gives your assistant the upgraded ability to complete any question or assignment with greater precision and personalization, tailored to however you like to work.

Ask Perplexity queries like: 

• “Recommend new running shoes for me.” (The assistant knows whether you’re injured, training for a marathon, or anything between)

• “I’m traveling to NYC next week. Suggest a good book for me to read on my flight.”

• “What gifts should I buy for my mom and brother for the holidays?”

• “What was the advice you gave me last week on dealing with my difficult coworker?

The responses reflect a user’s unique history and taste, synthesized from recent conversations and stored preferences to provide the most accurate answer possible. 

Stay in control 

Memory is a powerful addition to AI assistants, but there are exceptions where it’s helpful to have more control. You have complete control over what Perplexity remembers on your behalf: you can easily turn it off when needed. For added peace of mind, memory and your search history are automatically disabled in incognito mode, so nothing is saved while you browse privately. All data is encrypted, and you can even opt out of contributing to model improvement via the AI Data Retention settings. 

Calibrated context for user choice 

Perplexity remembers context across every model we offer. Users are free to choose whichever model they prefer, and change models from question to question. Unlike other platforms where your history lives within their system, with Perplexity you can carry your context across any model, choosing the best assistant for each task. 

Context portability is more than convenience and personalization, it’s a powerful way to protect the work you’ve already done. Hours spent building unique memory are still valuable when a new model launches. Users can switch between reasoning models for complex analysis, fast models for quick queries, and specialized models for domain-specific work, all working from the same personalization layer that becomes more nuanced the more you use it. 

The power of memory lies in knowing what to remember and when it matters. Perplexity transforms scattered context into coherent intelligence, and the Comet Assistant evolves alongside how you think and work.